Medical data mining for discovering periodically frequent diseases from transactional databases

Abstract

Medical data mining has witnessed significant progress in the recent past. It unearths the latent relationships among clinical attributes for finding interesting facts which helps experts in health care in decision making. Recently, frequent patterns in transactional medical databases that occur periodically are exploited to know the temporal aspects of various diseases. In this paper we modified K-means algorithm to extract yearly and monthly periodic frequent patterns from medical datasets. The datasets contain electronic health records of 2012 and 2013. Periodical frequent patterns between these years and monthly patterns were extracted using the proposed methodology. To achieve this we used the notion of making temporal view that is instrumental in adapting K-means for this purpose. We built a prototype to test the algorithm and the empirical results reveal that the proposed methodology for knowledge discovery related periodic frequent diseases is useful. The application can be reused to have lasting implications on health care industry for improving quality of services with strategic and expert decision making.